Chapter 9. The H-2 Genetic Complex, Dexamethasone-Induced Cleft Palate, and Other Craniofacial Anomalies

CHAPTER 9

THE H-2 GENETIC COMPLEX, DEXAMETHASONE-INDUCED CLEFT PALATE, AND OTHER CRANIOFACIAL ANOMALIES Joseph J . Bonner DENTAL RESEARCH INSTITUTE CENTER FOR THE HEALTH SCIENCES UNIVERSITY OF CALIFORNIA LOS ANGELES, CALIFORNIA

I. Introduction ..................... ......................... 11. H-2 and Chromosome 17-Backgro ......................... A. The Nobel Prize . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. The H-2 Gene Products.. . . . . . . . . . . . . . . . . . . . . . C. Congenic Strains of Mice.. . . . . . . . . . .................... 111. Linkage Analysis .............................................

193 194

196 198

D. The Grandmother Effect.. . B. C. D. E.

Dose-Response Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The kid Recombinants-Two Dcp Loci.. . . The aib Recombinants-Three l k p Lo Sex-Associated Gene . . . . . . . . . . . . . . . .

201

A. B. C. D. E.

Dexamethasone Has No Effect on EX, MG, and M I . . . . . . . . . . . EX in C-S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MG in C-S and D-&a-I. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MI in E,-Qa-I .................... Common Features of EX MG, and MI.. .....................

209 210 211

........................

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

211 214 214

1. Introduction

Results of recent experiments will expand the original observation that a gene associated with the major histocompatibility complex (H-2) influences susceptibility to glucocorticoid-induced cleft palate in the mouse (Bonner and Slavkin, 1975). The gene is Dcp (for dexametha193 CURRENT TOPICS IN DEVELOPMENTAL BIOLOGY, VOL 19

Copyright 0 1984 by Academic Press, Inc All rights of reproduction In any form reserved ISBN 0-12-153119-8

194

JOSEPH J. BONNER

sone-induced cleft palate, DCP). The experiments were intended to demonstrate linkage with a backcross text and to map the chromosomal position of the gene. The gene products of H-2 serve as the markers of subregions in chromosome 17. The results show the following: (1) H-2 and dexamethasone-induced cleft palate (Dcp)genes are linked. (2) Gene mapping analyses of H-2 genotypes that are recombinants of H-2k and H-2d alleles show there are two Dcp loci, each controlling a different mechanism. (3) Mapping analyses with H-2" and H-2b recombinants can be interpreted to show two or three Dcp loci. (4)Whether there are two or three alleles of each locus remains obscure. ( 5 ) There appears to be a complexity of interactions between the loci and alleles in the form of epistasis and/or complementation. (6) There is an interaction with a sex-associated gene. (7) The expanded linkage analysis suggests that a factor of DCP susceptibility is transmitted horizontally from mother to female offspring. Several unplanned observations come to light. First, genes associated with H-2 modulate the spontaneous frequency of craniofacial anomalies. These birth defects are exencephaly, micrognathia, and microphthalmia. Second, the occurrence of these craniofacial defects is not due to the teratogenic action of dexamethasone. Third, H-2-associated genes modulate the frequency of dorsoventral vaginal septa, a birth defect of the female reproductive tract. II. H-2 and Chromosome 17-Background

A. THENOBELPRIZE H-2 was discovered by George Snell, who won the Nobel Prize in 1981 for his discovery (Snell, 1981). H-2 quickly gained a prominent position in biological research because most regulatory functions of the immune response and transplant rejection are linked to it. The genes in H-2 seem to determine one's ability t o fight off infectious diseases, to keep autoimmunity in check, and to destroy virally infected cancer cells (Snell et aZ., 1976).

B. THEH-2 GENE PRODUCTS Three classes of molecules are encoded in H-2. Their detailed protein structure and gene structure are being described (Pease et al., 1982; Steinmetz et al., 1982; Klein et al. 19Sl>,and the distribution of their polymorphic alleles throughout natural populations is being investigated (Nadeau et al., 1982). These molecules now serve as definitive chromosomal markers in gene mapping, replacing the more subjective use of antigenic specificities.

9.

THE

H-2 COMPLEX

AND CLEFT PALATE

195

Class I molecules (K, D, etc., see Fig. 1) are cell surface glycoproteins found on many cell types (Klein et al., 1981).The quantity varies among cell types (Klein, 1975). The gene and protein structures of class I molecules are like that of an immunoglobulin (Pease et al., 1982). Class I1 molecules, also immunoglobulin-like in structure, are more restricted in their distribution, generally found on lymphoid cells and selected epithelial cells (Steinmetz et al., 1982).Class I11 molecules are serum components of the complement system (Roos et al, 1978). This system is activated by the antigen-antibody reactions. The function of class I and class IJ molecules appears to be antigen presentation and recognition by lymphocytes. Class I molecules are believed to be the markers of self, protecting the cell from autoimmune attack. The deviations from self expressed on the surfaces of virally infected cells are recognized by lymphocytes that have the same class I molecules as the infected cells. This is a characteristic called H-2 re-

CHROMOSOME 17 C cM

(H-2) Fu KD

T/t

5

7

5

C3 11

GT-2

T d

11

(Qa-2,Tla.Qa -1.Pgk 2)

cM

0.5

1.5

9

FIG.1. Illustration of chromosome 17 in the mouse. Each bar represents increasing detail of chromosomal structure from the centromere ( C )to the telomere (TI. Gene loci are marked and the distance in centimorgans (cM) between them is shown. The gene loci , vertebrae and tail abnormalities ( F u ) , in the top bar are the tail-affecting loci ( T i t ) fused H-2, component of complement ( C 3 ) ,and teratocarcinoma graft rejection (Gt-2). The middle bar focuses on the region between K and D, the classical transplantation antigens. Qa-I and &a-2 are newly described transplantation antigens expressed on lymphocytes, TZa is thymus leukemia antigen in normal thymus cells and some leukemia cells, Pgk-2 is phosphoglycerate kinase-2, and Upg-l is urinary pepsinogen-1. The bottom bar focuses on the H - 2 complex and the genes for class I molecules (KJI,Qa,TZa),class I1 molecules (A,,A,,E,,E,), and class I11 molecules C4, which is a component of complement, and S , which is the sex-limited serum protein whose function is unknown. The parentheses indicate that the order of the genes is not definitive. (Data taken from D. Klein et aZ., 1982, and Steinmetz el al., 1982.)

196

JOSEPH J. BONNER

striction (Zinkernagel and Doherty, 1975; J. Klein et al., 1982). Several investigators have proposed that the class I molecules, in addition to their role in the immune recognition system, play a role in intercellular recognition in embryonic development (Bennet et al., 1972; Boyse and Cantor, 1978; Bonner, 1979; Snell, 1980). As attractive as this hypothesis may be, there is no direct evidence that supports it. Nevertheless, the T complex, which affects the occurrence of early embryonic defects and tailless mice (Bennett, 1980; Artzt et al., 19821, and the H-2 complex, which affects the occurrence of craniofacial defects (see Section V), and the linkage between these two large gene complexes in chromosome 17 present an exciting research problem in mammalian genetics. Understanding their relationship will deepen our knowledge of regulatory mechanisms in intercellular communication and embryonic development. C. CONGENIC STRAINS OF MICE We can benefit now from the use of the exceptionally refined gene mapping of chromosome 17 made possible by the formation of 23-2 congenic strains of mice by Snell, Stimpfling, Shreffler, Klein, and Boyse. These inbred mice were selectively bred to have genetic differences restricted to only chromosome 17. The rest of the mouse genome is identical in every way among the strains in a congenic line. The mice used in our investigations are listed in Table I. All strains have the C57BL/10 genomic background and each has a different H-2 haplotype. The haplotype is the sequence of alleles of linked loci in the H-2 complex. For example, H-2b has mostly b alleles that are linked. TABLE I LIST OF CONCENICSTRAINS OF MICE AND THE H-2 HAPLOTYPE Strain C57BL/ lOScSn BlO.A/SgSn BlO.BR/SgSn BlO.D2/nSn BlO.A(2R)/SgSn BlO.A(4R)/Sg BlO.AI5R)ISgSn BlO.A(l8R)ISg

H-2 haplotype b a

k d h2 h4 i5 i18

9.

THE

H-2 COMPLEX AND CLEFT PALATE

197

H-2a, on the other hand, is a recombinant haplotype of two different linked alleles, k and d (see Table 11). In many cases the genetic difference between congenic strains extends beyond the traditionally defined H-2 complex from K to D (see Fig. 1).More often, large chromosomal segments are different between the strains and in some cases the whole chromosome may be different (D. Klein et al., 1982). The minimum genetic difference between strains C57BL/lO-H-Zb and BlO.A-H-2", for example, is K to &a-I; the maximum possible genetic difference may be from the centromere to Pgk-2. A gene for a phenotypic difference between these two strains probably maps in this subregion of the chromosome. A mutation elsewhere in the genome that may have occurred since the formation of the strains can cause the phenotypic difference too. Therefore linkage should always be confirmed with a backcross test, that is, in our case, a cross between DCP-resistant and -susceptible strains, then backcrossing the progeny with one of the parents, and finally, monitoring both the frequency of DCP and the H-2 haplotypes in the backcross progeny.

TABLE I1 LIST OF ALLELES FOR EACHOF THE H-2 HAP LO TYPES^ Alleles

H-2 haplotype

C

K

A

E,3

E,,

S

D

Qa-2

Tla

Qa-1

Pgk-2

Upg-2

T

h4

aorb

k

k

klb

b

b

b

a

b

b

U

S

b

h2

a o r b k

k

k

k

d

b

a

b

b

a

s

b

a

a o r b k k ..- - - -. b b b

k

k

d

d

a

a

a

a

S

b

b l k k d d

a

a

a

U

s

b

a

a

a

0.

S

b

a

S

b

a

S

b

i 5 i18 b

b

b

b

b

b

b

b

b

b

b

h

a

b

b

d

d

d

d

d

tl

a

c

b

--dorb

b

b

d

..-.

d

... .-.

k

k o r b k

- -. - - -.

_._.___._..___

k

k

k

k

k

b

a

a

a

s

korb

aHorizontal bars highlight allelic differences between adjacent H-2 haplotypes. Broken bars signify regions that may be different. The sequence of alleles in i18 is extrapolated from the data of D. Klein et al. (1982).C represents the centromere and T represents the telomere.

198

JOSEPH J. BONNER

111. Linkage Analysis

A. H-2 AND Dcp ARE LINKED

F, hybrids of C57BL/10 and BIO.A were backcrossed to C57BL/10 (Bonner and Tyan, 1982,1983b3. The backcross progeny were H-Zalb or H-Zblb. Forty-four pregnant females were given a dose of dexamethasone (a synthetic, high-potency glucocorticoid)on day 12 of pregnancy. On day 18 each fetus was scored for the presence or absence of cleft palate and then for its H-2 haplotype. The haplotype was determined with the immunofluorescent technique and monoclonal antibodies reactive to H-2Kk. Results of this backcross test are shown in Table 111. The frequency of DCP in the H-2"lb fetuses was 36%, and in H-2b/b fetuses, 23%. The statistical test for independence (the G statistic of Sokol and Rohlf, 1981) shows that H-2 and DCP susceptibility do not segregate independently (G = 5.47, p < 0.025). This experiment settles the issue of linkage. The difference in DCP susceptibility between C57BL/10 and BIO.A originates not from a gene elsewhere in the genome but from a gene linked to H-2. Not settled are the questions of the evolutionary origin of the genetic difference between the strains and the mechanism through which the gene is expressed. B. SURVIVAL ADVANTAGE FOR H-2 HETEROZYGOTES The rules for the Mendelian segregation of alleles predict that H-2"lb and H-2b/bfetuses occur in the backcross progeny in a 50:50 ratio. A significant deviation from this indicates a survival advantage for one of the two types. Table III does not contain an even number of homozygotes and heterozygotes. The ratio is 41:59 ( p < 0.0051, heterozygotes pre-

TABLE I11

THE2 x 2 TABLEFOR THE BACKCROSS TESTOF INDEPENDENT SEGREGATION OF H-2 AND Dcp H-2 haplotype ab bb

z

DCP 62 28 90

+

DCP 111

93 204

2 173 121 294

Percentage DCP 35.8 23.1

9.

THE

H-2 COMPLEX AND

CLEFT PALATE

199

dominating. There are two possible explanations. One is that H-2 heterozygotes have the survival advantage. This is not the first time an H-2 heterozygotic advantage has been reported, but it is not a consistent finding (Stimpfling and Richardson, 1965; Palm, 1974). The second explanation is that subjective bias favors positive fluorescence when scoring the fetal spleen cells in the immunofluorescence assay. AND EMBRYONIC FACTORS C. MATERNAL

How does the gene exert its influence? Maternal and embryonic factors are indicated. The H-2-linked Dcp gene exerts itself in both ways. In Table IV is the DCP frequency of reciprocal crosses between BIO.A (identified as A) and C57BL/10 (B) (Bonner and 'Tyan, 1983b). The difference between the two hybrids, the reciprocal crosses A ? x B d F, and B ? x A 6 F,, indicates that a maternal factor modulates the DCP frequency (see also the discussion by Vekemans and Biddle, this volume). The fetuses are genetically identical F , individuals, at least for chromosomal inheritance. Factors that could cause the differences are the maternal intrauterine environment and matriclinous inheritance. TABLE, IV

DCP FREQUENCY AT 160 MG/KG GIVENAS THE ARCSINE OF THE PERCE:NTAGE OF FETUSES WITH CLEFT PALATE I Y THE RECIPROCAL CROSSES A N D BACKCROSSES BETWEEN BIO.A AND C5713L/lQ

DCP

arcsine (SE)b

A x Ax B x B x Bx Bx (BIA) x (AIB) x

A B A B (BIA)c (AIB) B B

53.1 54.3 44.2 37.6 42.6 44.6 41.6 51.5

(2.2) (4.5) (4.0) (0.3) (4.4) (5.3) (3.8) (2.8)

"A, BIO.A; B, C57BLi10. 6 E , Standard error of the mean. 'BIA, C57BLi10 0 X BIO.A 6 F1; AIB, BIO.A P x C57BLi10 d F1.

200

JOSEPH J. BONNER

The difference in DCP frequency between the parental strain, B x B, and the cross, B x A, indicates that there is an embryonic factor as well, because the mothers in this comparison are the same genotype. Only the embryonic genotypes differ. This property of maternal and embryonic factors of DCP genetics has been observed by many investigators (Kalter, 1965; Bonner and Slavkin, 1975; Biddle and Fraser, 1976; Melnick et al., 1981).

D. THE GRANDMOTHER EFFECT The DCP frequencies of the backcrosses are listed also in Table IV. The H-2 genotypes of the dams in the backcrosses were interchanged between F, hybrid and homozygotic females. Additional variation in the F, hybrids was added so that dams and sires were either A x B F, or B x A F,. When the dam was B and the sire was A x B F, or B x A F, there was no significant difference between the DCP frequencies in the backcross progeny. When the sire was B and the dam was either A x B F, or B x A F, there was a significant difference in the DCP frequencies in the backcross progeny ( p < 0.05) (Bonner and Tyan, 1983b). The experiment was done with two doses of dexamethasone and in each case the A x B F, dam's progeny had a higher frequency of DCP. This maternal effect was imposed alike on H-2"Ib and H-ZbIb fetuses. The difference between the A x B F, and B x A I?, dams caused a reduction in the DCP frequency in both H-2"lb and H-ZbIb fetuses. What causes this difference? What is different between the A X B F, and B x A F, dams? It is not somatic or sex-linked genes, because the dams are genetically identical F, mice and they are congenic, with the same X chromosomes. It is not specific cytoplasmic inheritance, like that of mitochondria, because the breeding method to produce the congenic strains removes any possibility of matriclinous or patriclinous dissimilarity in the vertical transmission of inheritance (unless the A strain acquired it since the strain was made). A possibility that comes to mind is pre- or postnatal horizontal transmission of inheritance. A factor in the maternal intrauterine environment or in the sucklings' environment could be altering the F, dams. When the F, females became pregnant adults in the backcross test, whatever it was that altered them pre- or postnatally manifested itself by causing the difference in DCP susceptibility in their backcross progeny. In relation to the backcross progeny the difference is thus a "grandmother effect." Fetuses with H-2" grandmothers are more susceptible to DCP than fetuses with H-Zb grandmothers.

9.

THE

H-2 COMPLEX

AND CLEFT PALATE

201

IV. Gene Mapping Analyses

A. THE STRATEGY Now that linkage between H-2 and Dcp is established firmly, the next useful information is the location of the gene in chromosome 17. How many centimorgans from H-2 is it? The traditional method to measure the recombination frequency between two linked loci such as H-2 and Dcp is not appropriate, because DCP is a quantitative trait and to measure the shift in frequency resulting from recombination in backcrosses would be experimentally impractical. Instead, the known H-2 recombinant haplotypes can be used to subdivide chromosome 17 into subregions and measure the effect of each chromosomal subregion on DCP susceptibility. The best we can hope to do is to find which H-2 gene loci lie closest to Dcp. This must suffice until a gene product for DCP is discovered. Only DCP susceptibility differences between strains give useful information. No difference means no information, but DCP susceptibility similarities do not mean that two strains have the same gene in regions of genetic similarity. Only physical properties of a gene or its product show that two strains have the same gene. B. DOSE-RESPONSE ANALYSIS An experimental method for measuring DCP susceptibility with a sensitive ability to detect strain differences is dose-response analysis. Several measurements are made simultaneously; each can be used to detect strain differences. The measurements are (1)the slope of the line (the regression coefficient); (2) t h e y intercept or some other point on the y axis that corresponds to a dose on the z axis; and (3) the variance, which is a measure of the amount of variation in the dose response of DCP. Different slopes of DCP dose-response lines indicate that there are different mechanisms through which dexamethasone induces the cleft palate. Differences between the y intercepts or y coordinates at specified doses indicate that the strains differ in sensitivity to dexamethasone. When the strains differ in variance, one strain has a larger range of responses to dexamethasone than the other. Results are summarized in Table V and Figs. 2 and 3. C. THE kld RECOMBINANTS-TWODcp LOCI The first strain comparisons ascertain the effect of two H-2 subregions. The strains are BlO.A-H-2", B10.D2-H-2d, and B10.BR-H-2k.

202

JOSEPH J. BONNER

TABLE V RESULTSOF

H-2 haplotype b

i5 a

h2 il8 k d

THE

DOSE-RESPONSE ANALYSIS OF SEVEN CONGENIC STRAINS WITH DIFFERENT H - 2 HAPLOTYPES Regression coefficient (SEP 37.5 45.3 45.3 51.9 54.5 67.3 70.5

Estimated arcsine a t 160 mgikg

Variance

37.6 36.9 53.1 54.5 39.7 33.1 25.0

406 332 283 295 283 273 344

(1.4) (4.8) (4.1) (5.5) (6.0) (10.8) (9.7)

aSE, Standard error of the mean.

The H-2" haplotype is a recombinant of both k and d alleles (see Table 11). The minimal genetic difference between H-2" and H-2d is K-E, and Tla-Qa-l . The minimal genetic difference between H-2" and H-2k is S-&a-2. These strains differ in DCP susceptibility ( p < 0.001) (Bonner and Tyan, 1983a). They also differ in regression coefficients ( p < 0.05). The difference in DCP susceptibility can be seen in Table V by comparing the estimated cleft palate arcsine at 160 mg/kg. H-2" is the

I

1.5

2.1

2.7

Log Dose Dexamethasone FIG. 2. Dexamethasone-induced cleft palate dose-response analyses of three H - 2 congenic strains of mice. On t h e y axis is the arcsine, which represents the frequency of cleft palate. An arcsine of 45 is equivalent to 50% cleft palate. The x axis is the log of the doses of dexamethasone given to pregnant mice on day 12 of pregnancy (log 2.2 is equivalent to 160 mgikg dexamethasone). The results are of strains B10.AISgSn (a), C57BL/10Sn (b), and B10.BRISgSn (k).

9.

THE

2oj -b

H-2 COMPLEX

1.5

AND CLEFT PALATE

I

2.1

I

203

I

2.7

Log Dose Dexamethasone FIG. 3. Dexamethasone-induced cleft palate dose-response analyses of four additional H-2 congenic strains. The strains are BlO.A(2R)/SgSn (hZ), BlO.A(l8R)ISg (i18), BlO.A(5R)/SgSn (i5), and B1O.DZinSn (d).

susceptible haplotype, and H-2k and H-2" are resistant. H-2" has a regression coefficient that is less than either H-2k or H-2d. The difference in the regression coefficients suggests that there are two different mechanisms through which dexamethasone induces cleft palate. To substantiate the two mechanisms, a cross of the strains in all possible combinations detected maternal and embryonic effects. The data are shown in Table VI and Fig. 4 (Bonner and Tyan, 1983a). Reciprocal crosses between H-2" and H-Zk show only an embryonic effect. On the other hand, crosses between H-2" and H-2d show both maternal and embryonic effects. H-.Zk and H-Zd crosses show no differences except that H-Zk is slightly more susceptible than H-2d. The kld recombinant haplotypes analyses suggest two Dcp loci, each with a different mechanism. The K-E, and/or Tla-Qa-1 chromosomal subregion affect both maternal and embryonic factors. The SQa-2 subregion affects only an embryonic factor. I carefully choose the word "suggest" because there is some doubt. Uncharted chromosomal regions could be different between these strains (D. Klein et al., 1982). The centromeric and telomeric regions in particular could be k or b for H-2", d or b for H-2", and k or b for H-2k (see Table 11). These uncharted chromosomal regions leave an element of doubt and uncertainty in mapping studies.

D. THEalb RECOMBINANTS-THREE Dcp LOCI The alb recombinants are more suitable for mapping studies because doubt from uncharted chromosomal regions is mostly elimi-

204

JOSEPH J. BONNER

TABLE VI DCP FREQUENCY GIVENAS THE ARCSINEOF THE PERCENTAGE OF FETUSES WITH CLEFTPALATE IN THE RECIPROCAL CROSSESBETWEEN BlO.A, B10.D2, A N D B1O.BR Dose (mgikg) 160 160 160 160 160 160 160 160 226 226 226 226

Crossa 9 x 6 A A K K A A D D K K D D

x x x x x x x x x x x x

A K A K A D A D K D K D

DCP arcsine (SEP 53.1 47.4 51.7 33.1 53.1 39.8 29.7 25.0 43.4 36.1 38.1 35.6

(2.2) (3.1) (4.3) (2.3) (2.2) (5.4) (3.8) (2.7) (2.3) (5.0) (6.2) (1.8)

"A, B1O.A; D, B10.D2; K, B1O.BR. bSE, Standard error of the mean.

nated. For example, H-2h4, H-2h2, and H-2" have the same chromosomal region from the centromere to K, whether it is a or b. Likewise, H-2i5, H-2i18, and H-2b have the same region from the centromere to K, which is b. All alb recombinant congenic strains have the same chromosomal region from Pgk-2 to the telomere.

1. Dcp-1 in C-E, The significant difference in DCP susceptibility between H-2" and H-P5 ( p < 0.001) can be seen by comparing the cleft palate arcsine at 160 mg/kg in Table V (Bonner and Tyan, 1983b). It indicates a Dcp locus somewhere between the centromere (C) and E,, a chromosomal distance of approximately 15 cM (D. Klein et al., 1982). The end point of this region is in the structural gene for E,, a class I1 molecule. The point of recombination in E , was identified by Steinmetz et al. (1982). A more precise chromosomal position (less than 15 cM) of this locus, tentatively designated Dcp-1, cannot be identified with these congenic strains until markers between C and K are found. 2. Dcp-2 in E,-S

The second locus, Dcp-2, is in a 0.2-cM chromosomal distance marked by E , and S. The effect of this locus can be seen in two ways by

9.

T H E H-2 COMPLEX AND CLEFT PALATE

p---+.,J

205

EMBRYONIC EFFECT

‘x

!?i 20

kd/kd

kd/kk

kk/kd

kk/kk

HAPLOTYPE

HAPLOTYPE

FIG. 4. A graphic representation of the results of the reciprocal cross test with the cleft palate frequency expressed as a n arcsine on t h e y axis and the H-2 haplotype of the cross on the x axis. The length of the bar is t standard error of the mean. Crosses are (a) BIO.A (kd) and B1O.BR (kk): (b) BIO.A (kd) and B10.D2 (dd); (c) blO.BR (kk) and B1O.D2 (dd). (Reprinted with permission from Bonner and Tyan, 1983a.)

incorporating the data presented here with data from a report by Gasser et al. (1981) (see also discussion by Goldman in this volume). The comparison between H-2i5 and H-2lI8 shows a Dcp locus with a weak effect on susceptibility in E,-S. The effect on susceptibility was 36.939.7 ( p < 0.05) cleft palate arcsine a t 160 mg/kg. The allelic change was EE-Sd to Ek-Sb. In Gasser’s report a single dose of cortisone was used to induce cleft palate. A first approximation of susceptibility is possible with this technique. They reported the same pattern of suscep-

206

JOSEPH J. BONNER

tibility that we saw with the dose-response analyses. That is, H-2" and H-2h2are susceptible; H-2b,H-F5, H-2k, and H-2d are resistant. It was The additional allele of interest that they tested was found that H-2h4was indistinguishable from H-2h. The sample size was 10 litters, so a dose-response analysis should confirm the finding. A safe assumption is that the magnitude of the difference between H-2h2and H-2h4is comparable to what we saw between H-2" and H-Zi5. The genetic difference between H P 4 and H-2h2 is E,-S. The allelic change is E$-Sd to E;-Sb, and it had a strong effect on DCP susceptibility. Why in one case does the allelic change in the ED-S subregion have a weak effect (that is, the i5 and i18 comparison) and in another a strong effect (the h2 and h4 Comparison)? There are several possibilities. First, the point of recombination in i5 and h4 is separated by approximately 5000 nucleotide bases (Steinmetz et al., 1982). A possibility is that the structural difference between the two E , genes caused by recombination caused different magnitudes of change in DCP susceptibility. This possibility directly implicates E , as the Dcp gene. Second, if Dcp-1 is not E , and the point of recombination between h2 and i18 in the subregion between S and D is the same, the data then suggest an epistatic interaction between Dcp-1 in C-E, and Dcp-2 in E,-S. The K allele of C-E, allows expression of the allelic change at E,-S, but the b allele of C-E, suppresses the expression of the allelic change of E,-S. Third, assuming that the point of recombination between S and D of haplotypes h2 and 218 is not the same, the point of recombination in h2 may be closer t o D and it may include a third Dcp locus. In this case the comparison between h2 and h4 shows the additive effect of allelic changes in Dcp-2 and Dcp-3, but the i5 and i18 comparison shows the effect of an allelic change at Dcp-2.

3. Dcp-3 in D-Qa-1 The arguments for epistatic interactions and different points of recombination between h2 and i18 apply to the proposal for a third Dcp locus in the D-Qa-1 subregion. It is a chromosomal distance of 1.7 cM. Evidence for Dcp-3 lies in the comparison between H-2band H-2i18. There is a difference between the strains' susceptibility and variance in this comparison (see Table V). The allelic change is Db-Qa-lb to Dd-Qa-1". Again we have a paradox; this allelic change is the same as in the H-2" and H-2hZcomparison. There is no phenotypic difference between H-2" and H-2h2( p > 0.25) (Bonner and Tyan, 1983b).If points of recombination between h2 and i18 are the same, then epistasis applies. But if the crossover of h2 is different from that of 218, then

9.

THE

207

H-2 COMPLEX AND CLEFT PALATE

H-2" and H-2m may have the same allele at Dcp-3 and H-2b and H-2zl8 may have different Dcp-3 alleles. These arguments will be resolved only with the nucleotide sequences of the S and D subregions that show the points of recombination. 4 . Summary

One interpretation of the mapping analyses of the genes linked to H-2 that affect dexamethasone-induced cleft palate is that there are three loci. The loci are tentatively designated Dcp. Dcp-I is in the chromosomal region marked by C-E,; Dcp-2 is in the region E,-S; Dcp-3 is in the subregion D-&a-I kee Fig. 5 ) . Three alleles, H-2b,H-2k, and H-Zd,were observed, but the overlap between the recombinants and the uncharted chromosomal subregions precludes distinguishing between the two or three alleles of each Dcp locus. Another interpretation of the data on alb recombinants is that there are two Dcp loci. Implicit in this interpretation is that the Dcp genes are in the regions of overlap i n the analysis of the alb recombinants. The Dcp-I locus is gene E,. Dcp-2 is in the region between S and D (see Fig. 5). This conclusion yields a refined map indeed.

E. SEX-ASSOCIATED GENE Francis (1973) reported that sex-linked genes modify susceptibility to glucocorticoid-induced cleft palate in the mouse. In some inbred

CHROMOSOME 17 1

d

". 2

EUS D

3 Oa-l

Pgk-2

T

FIG.5 . Graphic representation of the two interpretations of the gene mapping data of the H-2-linked dexamethasone-induced cleft palate susceptibility genes (Dcp i n the text). Across the top are (1)Dcp-l in the 15-cM chromosomal segment from the centromere (C)toE,; (2)Dcp-2 in the 0.2-cM segment from E , toD; and (3)Dcp-3 in the 1.7cM segment from D to &a-I. The alternative interpretation is that genes are in the regions of chromosomal overlap when comparing the aib recombinant haplotypes. Overlap is marked by the 1 and 2 and the arrows on the bottom of the illustration; in this case Dcp-1 is the structural gene for E , and Dcp-2 is in the chromosomal segment between S and D .

208

JOSEPH J. BONNER

strains the females have a higher frequency of cortisone-induced cleft palate than males. However, surveys of inbred and congenic strains indicated that expression of high female sensitivity is variable (Loevy, 1972, Biddle and Fraser, 1976; Tyan and Miller, 1978). The sex of many fetuses in our investigations was determined by internal examination of the gonads. The sex distribution in the fetuses and the results of the G test (Sokol and Rohlf, 1981) for the independent occurrence of sex and cleft palate are shown in Table VII. The only strain with a statistically significant association between sex and DCP was BlO.A(l8R). Next was BlO.A(5Rj but the level of confidence was less than 90%. This observation suggests there is an interaction between the Dcp loci linked to H-2 and a gene associated with sex. The combination of alleles in the H-2i1Rhaplotype allows expression of the sex-associated effect. All others, except H-,F5, hide its expression. The interaction may be epistasis. V. H-2 and Craniofacial Anomalies

During experiments to map the Dcp genes linked t o H-2, the frequency of other gross craniofacial anomalies was recorded. Two unplanned observations are notable. First, genes associated with H-2 haplotypes influence the frequency of occurrence of exencephaly (EX, incomplete anterior neural tube formation), micrognathia (MG, incomplete jaw formation), and microphthalmia (MI, incomplete eye formation j. The second notable observation was that dexamethasone’s

TABLE VII

DCP FREQUENCY IN MALESAND FEMALES OF SIXCONGENIC STRAINS OF MICE H-2 haplotype i18

h5 b

a

k

h2

Percentage DCP in females

Percentage DCP in males

G statistic

49 41 50 33 45 42

40 34 47 32 46 44

6.4, p < 0.05 3.0, p < 0.1 0.2, nsa 0.1, ns 0.1, ns 0.3, ns

ans, Not significant.

9.

THE

H-2 COMPLEX AND CLEFT PALATE

209

teratogenic action had no effect on the frequency of these craniofacial birth defects. An analysis of the alb recombinant haplotypes shows the subregion of chromosome 17 that contains genes affecting the frequency of each anomaly. The frequency of EX is influenced by C-S, MG by C-S and D-&a-I, and MI by Ep-Qa-l. There also seem to be chromosomal interactions between subregions. Other subregions are implicated by the comparisons of the kld recombinant haplotypes, but the uncertainty of the chromosomal differences between these strains preclude serious discussion of the mapping possibilities. A. DEXAMETHASONE HASNO EFFECTON EX, MG,

AND

MI

Table VIII shows the comparison between increasing doses of dexamethasone and the frequency of occurrence on MI, MG, and EX in strain C57BL/10. There is a considerable amount of variation in the arcsine percent (5%) frequency of MI and MG. This number is the number of litters with defective fetuses (the litter is the responding unit) in proportion to the total number of litters in the group. In one group (280 mg/kg) no fetus had MI out of 11 litters, while in another group 6 litters out of 31 had a fetus with MI. A regression analysis was done with a single value of y for each value of x (Sokol and Rohlf, 1981) TABLE: VIII DOSE-RESPONSEANALYSIS:DEXAMETHASONE AND THE FREQUENCY OF MICROPHTHALMIA, MICROGNATHIA, A N D EXENCEPHALY Number of litters with defective fetusesa Dose (mgkg) 0 80 120 155 217 280 340

Total

Total litters

MI

20 15 11 31 30 11 18

1 (15.3) 3 (27.8) 1 (21.0) 6 (26.8) 5 (24.9) 0 (8.4) 3 (25.4)

136

19

MG

=

0.27, p > 0.5.

0 0 0 1 1 0 0

1 (15.3) 1 (17.6) 5 (42.6) 4 (24.5) 4 (22.4) 1 (20.4) 1 (21.2)

17

aArcsine values in parentheses. For MI, F (1,4)

p > 0.25. For MG, F (1,4)

EX

2 =

0.83,

210

JOSEPH J. BONNER

and the results are shown at the bottom of Table VIII. There was no significant regression of the MI or MG frequency on the dose of dexamethasone ( p > 0.25 and p > 0.5). EX was too infrequent to be analyzed, but when one considers all strains and all doses there was no significant regression. This same pattern was seen for seven congenic strains (Bonner et al., 1983). It can be stated with a high degree of confidence that dexamethasone does not influence the frequency of MI, MG, and EX and that our observation is the spontaneous occurrence of these anomalies. At times the anomalies appeared t o cluster in incidence in time, but the experiments were not planned with time in mind and a planned observation should be made. B. E X I N C - S EX occurred the least number of times (see Table 1x1. Only one fetus per litter ever had the anomaly. The frequency of occurrence among the congenic strains with different H-2 haplotypes ranged from 0.5% for H-2i18 to 5.5% for H-2a. There is significant heterogeneity among the frequencies for the strains ( p < 0.025). The results of the unplanned tests of the homogeneity of replicates tested for goodness of fit using the G statistic (Sokol and Rohlf, 1981) are shown at the bottom of Table IX. Comparisons of the alb recombinant haplotypes indicate the chromosomal subregions that are affecting the frequency of EX. Comparing the H-2" frequency and H-2i18frequency, the difference between TABLE IX THE FREQUENCY OF OCCURRENCE OF EXENCEPHALY, MICROGNATHIA, AND MICROPHTHALMIA IN SEVEN H-2 CONGENIC STRAINS OF MICE Number of litters with defective fetusesa

H-2

haplotype k

i5 a

i18 h2 b d

Total litters

MI

MG

EX

122 112 181 215 128 136 84

31 (25.4) 29 124.0) 39 (21.5) 36 (16.7) 20 (15.6) 19 (14.0) 1 (1.2)

10 (8.2) 6 (5.4) 13 (7.2) 5 (2.3) 8 (6.3) 17 (12.5) 1 (1.2)

1 (0.8) 3 (2.7) 10 (5.5) 1 (0.5) 2 (1.6) 2 (1.5) 3 (3.6)

OPercentage values in parentheses. For MI, G (6) = 37.8, p < 0.001. For MG, G (6) = 21.0, p < 0.005. For EX, G (6) = 13.8, p < 0.025.

9.

THE

H-2 COMPLEX AND CLEFT

PALATE

211

0.5 and 5.5%is significant ( p < 0.005). The maximal genetic difference between these haplotypes is from the centromere to the S subregion of H-2. There appears t o be an epistatic interaction between two loci because the genetic difference between H-Zb and H-2h2 is the same chromosomal region, but there is no difference of EX frequency between them.

c. MG IN c-s AND D-Qa-1 MG occurred in frequencies that ranged from 1.2% 0 f H - 2litters ~ to 12.5% of H-Zb litters. There is significant heterogeneity among the frequencies among the strains ( p < 0.005) (see Table 1x1. There are two strain comparisons that are appropriate for the gene mapping analysis. They are H-2b compared to H-2218,and H-2" compared t o H-2"IS. The genetic difference between H-2b and H-2lI8 is the D-Qa-1 subregion and the difference between 12.5 and 2.3%is significant ( p < 0.001). The maximal genetic difference between H-2" and H-2'18 is C-S, and the difference between 2.3 and 7.2%is significant ( p < 0.025). There is no overlap between the two chromosomal subregions, so a conclusion is that two genes affect the MG frequency. Epistasis applies to the frequency of MG because there is a difference between H-2b and H-2118 but no difference between H-2" and H-2&. Both comparisons, however, have the same genetic difference.

D. MI IN E,-Qa-1 MI is the most frequently occurring anomaly in these congenic strains of mice. Often it occurs twice in the same litter (Bonner et al., 1983). Approximately 26% of H-2k litters had the anomaly and only 1.2%of H-2d litters did. There are strain-associated differences in the frequencies ( p < 0.001). The only a / b recombinant comparison with a statistically significant difference is H-215 and H-2b, 24 and 14 %, respectively ( p < 0.025). The genetic difference between them is the Ep-Qa-l subregion. OF EX, MG, AND MI E. COMMONFEATURES

These craniofacial anomalies have common features. First, they usually occur in females. Second, sometimes two of the defects will occur together in the same fetus, MI and MG in particular. Third, the birth defects are in organs that are components, extensions, or induced derivatives of the anterior neural tube. Fourth, they occur in a wide range of severity and sometimes border on normality. Fifth, each results from incomplete morphogenesis and growth. Sixth, they do not

212

JOSEPH J. BONNER

breed true, at least for MI, and this is probably the case for MG and EX too. EX and usually MG are lethal; neonates die, so a test of inheritance is impossible. Mice with MI usually survive and if they breed, the frequency of MI in their progeny is no different from the inbred strain (Dagg, 1966). MI mice are phenodeviants described in the C57BL/6 and C57BL/10 strains and the occurrence of MI is probably induced by an environmental factor (Dagg, 1966). One impression is that EX, MG, and MI do not result from mutant genes but are the products of normal genes. The normal genes control morphogenesis, size, and shape of the developing organs and the environment causes variations in the expression of these genes. Mice with these birth defects probably are in the extreme end of a probability distribution graph of variation in size and shape. The genes associated with the H-2 haplotypes cause fluctuations in the shape of the curve so that one strain rarely has a fetus in the MG portion of the graph, for example, while another strain often has one in the “small jaw” end of the graph.

F. A PUZZLING FEATURE At first glance one would think that these three birth defects are actually varying degrees of severity of the same defect, EX as the most severe case and MI as the least severe. If that were the situation, one would predict that the rank order of the strains would be the same for the defects. In fact the rank order of the strains is not the same for each. MI and MG rarely occur in H-2d litters, yet H-2d ranks second in the frequency of EX. Another example is H-2b. It has the highest frequency of MG but it ranks close to the bottom of the rank for both MI and EX. Even the observation that the defects sometimes occur together in the same fetus should not mislead us because the occurrence of multiple defects may be a function of fetal liability and susceptibility and not a common genetic mechanism for the three birth defects. It must be stressed that these data demonstrate an association between H-2 haplotypes and the frequency of EX, MG, and MI. The data do not demonstrate linkage. A backcross test must be done to demonstrate that. Without the confirming backcross test, one possible interpretation of the data is that a significant amount of genetic and evolutionary divergence has occurred since these congenic strains were formed. The differences in phenotypes reported here may be the result of random mutations at many points in the genome. And if genetic drift occurs at such a rapid rate, the genetic foundation upon

9.

THE

H - 2 COMPLEX

213

AND CLEFT PALATE

which the concept of congenic strains of mice is built ought to be reevaluated. VI. H-2 and Dorsoventral Vaginal Septa

While probing a countless number of females for vaginal plugs for inclusion in the study on birth defects, we noticed that the frequency of dorsoventral vaginal septa (DVS) varied among the H-2 congenic strains. The data are shown in Table X (Bonner, 1981; Bonner and Tyan, 1983~).DVS occurred most frequently in females with H-zL5 haplotype and least frequently in H-2" females. Here, too, there are strain-associated differences in the frequencies ( p < 0.001). The mapping analysis with the aib recombinant haplotypes show that two subregions of H-2 affect the DVS frequency. The maximal genetic difference between H-2" and H-,P5 is C-E, and the difference in frequency of DVS is significant ( p < 0.001). The maximal genetic difference between H-2h2 and H-2" is D-&a-1 and the DVS frequency difference is significant ( p < 0.001). The concept of epistasis applies because the genetic difference between H-Zb and H-2118 is the same as for H-2" and H-2h2, yet there is no difference between the DVS frequencies of H-Zb and H-2"'". The frequency of DVS was also observed in the F, females of a cross between BIO.A and C57BLI10. There was no difference between the reciprocal crosses (Bonner, 19811, which indicated that there was no maternal effect as in the DCP frequency.

TABLE X

THEFREQUENCY OF DORSOVENTRAL VAGINALSEPTA IN SEVEN CONGENIC S F R A I N S OF MICEQ H-2 haplotype

Total females

DVS females

Percentage

i5 h2 b d i18 k

166 182

41 36 52 20 34 22 18

25 20 20 18 17 14 6

a

265 109 20 1 161 287

nG (6) = 38.9, p < 0.001.

214

JOSEPH J. BONNER

VII. Concluding Remark

Is the link between H - 2 and birth defects coincidental or functional? This is a question of fundamental interest. Functional is the more interesting answer. With this answer a theoretical framework could be built that would describe molecular processes used by the cell t o sense and respond to its environment. The theory adapts the immunoglobulin-like molecules of H-2 to roles as sensors on the cell surface. When the Ig-like molecules sense and bind to arrangements of extracellular molecules or molecules on the surfaces of adjacent cells, the cells respond by differentiating. This process is analogous to the process of antigen recognition by cell interactions in the immune response and lymphocyte differentiation, a glucocorticoid-sensitive process. It is not too far-fetched to believe that “Mother Nature” would use the molecular process of self-recognition of development as the molecular process for nonself recognition in the immune response. ACKNOWLEDGMENTS

I thank William Harris for editing and the UCLA Word Processing Center for manuscript preparation. The research was supported by Grant DE-05165 from the National Institute for Dental Research. REFERENCES Artzt, K., Shin, H. S., and Bennett, D. (1982). Cell 28, 471-476. Bennett, D. (1980). Huruey Lect. 74, 1-21. Bennett, D., Boyse, E. A., and Old, L. J. (1972). In “Cell Interactions” (L. G. Silvestri, ed.), pp. 247-263. American Elsevier, New York. Biddle, F. G., and Fraser, F. C. (1976). Genetics 84, 743-754. Bonner, J. J. (1979). Birth Defects: Orig. Article Ser. 15, 55-88. Bonner, J. J. (1981), J. Immunogenet. 8,455-458. Bonner, J. J., and Slavkin, H. C. (1975). Immunogenetics 2, 214-218. Bonner, J. J.,and Tyan, M. L. (1982). Teratology 26, 213-216. Bonner, J. J., and Tyan, M. L. (1983a). Genetics 103, 263-276. Bonner, J. J., and Tyan, M. L. (198310). In preparation. Bonner, J.J., and Tyan, M. L. (1983~). J.Zmmunogenet. (in press). Bonner, J. J., Dixon, A. D., Baumann, A., Riviere, G. R., and Tyan, M. L., (1983). In preparation. Boyse, E. A,, and Cantor, H. (1978). Birth Defects: Orig. Article Ser. 14, 249-269. Dagg, C. P. (1966).In “Biology of the Laboratory Mouse” (E. L. Greened.), pp. 309-328. McGraw-Hill, New York. Gasser, D. L., Mele, D., Lees, D. D., and Goldman, A. S. (1981). Proc. Null. Acud. Sci. U.S.A.78, 3147-3150. Francis, B. M. (1973). Teratology 7 , 119-126. Kalter, H. (1965).In “Teratology: Principles and Techniques” (J.G. Wilson and J. Warkany, eds.), pp. 57-79. Univ. of Chicago Press, Chicago.

9.

THE

H-2 COMPLEX AND CLEFT PALATE

215

Klein, J. (1975).‘:Biology of the Mouse Histocompatibility-2 Complex.” Springer-Verlag, Berlin and New York. Klein, J., Juretic, A,, Baxevanis, C. N., and Nagy, Z. A. (1981). Nature (London) 291, 455-460. Klein, J., Mursic, M., and Nagy, Z. A. (19821. Transplant. Proc. 16, 581-582. Klein, D., Tewarson, S. Figueroa, F., and Klein, J. (1982). Immunogenetics 16,319-328. Loevy, H. T. (1972). J . Dent. Res. 51, 1010-1014. Melnick, M., Jaskoll, T., and Slavkin, H. C. (1981). Immunogenetics 13,443-449. Nadeau, J., Collins, R. L., and Klein, J. (1982). Genetics 102, 583-596. Palm, J . (1974). Cancer Res. 34, 2061-2063. Pease, L. R., Nathenson, S. G., and Leinwand, L. A. (1982).Nature (London) 298, 382385. Roos, M. H., Atkinson, J. P., and Shreffler, I). C. (1978). J . Immunol. 121, 1106-1115. Snell, G. D. (1980).Harvey Lect. 74, 49-80. Snell, G. D. (1981). Science 213, 172-178. Snell, G. D., Dausset, J., and Nathenson, S. (1976). “Histocompatibility.” Academic Press, New York. Sokol, R. R., and Rohlf, F. J. (1981). “Biometry.” Freeman, San Francisco. Steinmetz, M.,Minard, K., Horvath, S.,McNicholas, J., Frelinger, J.,Wake, C., Long, E., Mach, B., and Hood, L. (1982). Nature (London) 300, 3b142. Stimpfling, J . H., and Richardson, A. (1965).Genetics 51, 831-846. Tyan, M. L., and Miller, K. K. (1978). Proc. Soc. Exp. Biol. Med. 158, 618-621. Zinernagel, R. M., and Doherty, P. C . (1975).J . Exp. Med. 141, 1427-1436.